19

A little bit of background first: Over on Puzzling.SE, we've been nuking users that look very spammy, before they've had a chance to post anything. It began with posts like this, which advertised watching things online and stuff.

So then we (I'm not a mod there, I just pinged them with links) started nuking users with spammy profiles before they had a chance to post. Like this 3D Printing user. They have keywords that might show up in a Google search and a link to their site.

Then the line gets a bit blurrier with profiles like this. The entire profile looks like an ad for the product, and they have no positive participation. The profile is still something I'd flag on the spot as spam.

...but now, after a whole long conversation in the Teacher's Lounge (link will only work for mods), it turns out that apparently we (me, the Puzzling mods, and apparently doppelgreener) are in the minority here. That most mods will not nuke a user based on that, and will wait for them to actually post spam, the concern being that a legit user will be thrown under the bus - with this person being an example of someone who might suffer from being nuked unfairly. (My opinion is that if they look like a spammer, they should expect to be treated like a spammer, and that if they want to be a positive participant they should create a less suspicious account.)

So... to resolve this (the conversation took over two hours), is there any official policy or anything on this?


Note: I'm aware of the Rules for Profiles. This is different, specifically as 1.) This is about the rules for nuking an account, and 2.) This has caused a lot of debate in the past couple hours and I feel it's time for a new discussion on this.

  • 7
    Advertising in the profile page is allowed, period. (unless the policy has changed, which I highly doubt. In such case a CM can post new answer on the dupe, or reopen this and post official answer.) – Shadow Jul 4 '17 at 14:00
  • 6
    @ShadowWizard - I'm trying to raise a new discussion on this. – user58 Jul 4 '17 at 14:01
  • 3
    No need, as in all other cases, old discussions can be bumped using proper actions. "Long time passed" is not valid reason to re-post something. – Shadow Jul 4 '17 at 14:02
  • 5
    @ShadowWizard - no, but 'is it time to revisit this' is. – user58 Jul 4 '17 at 14:03
  • 12
    To give some background to why this should be reopened (in both senses): the topic has caused a lot of discussion among moderators network-wide today, and we collectively felt that the policy is not clearly enough defined as things stand. – Dan Hulme Jul 4 '17 at 14:06
  • 2
    Then it needs to be clear what new facts need to be taken into account not addressed in the dupe yet. – rene Jul 4 '17 at 14:10
  • "My opinion is that if they look like a spammer, they should expect to be treated like a spammer" I'm afraid of what you would do to someone buying large knives, he/she may be preparing a murder after all... (kind of parabolic, but that's the idea) – Tensibai Jul 4 '17 at 14:38
  • @Tensibai I totally agree, and using the knifes metaphor, SE allows anyone to buy as many knifes as they want, and act only when someone use the knife to do something bad. – Shadow Jul 4 '17 at 14:46
  • @ShadowWizard actually, "long time passed" is a very good reason to repost on meta. Not on main sites, obviously, but I see no reason why a discussion that happened 6 years ago or whenever, back when we only had the trilogy, should never be repeated. Some posts don't age, yes, but others do and some meta discussions should happen again if the situation has changed. Again, not really talking about this one necessarily, just the general idea. – terdon Jul 4 '17 at 14:50
  • @terdon or closer to 8 years in this case :) – Cai Jul 4 '17 at 15:20
  • 1
    So THIS explains why Literature has so few users compared to other betas - Mithrandir keeps nuking them! :-P – Rand al'Thor Jul 4 '17 at 21:00
25

There isn't a clearly defined policy available because... we don't have one. We actually don't encourage moderators to seek out these kinds of users and destroy them because it's usually a waste of time. There are lots of spam users that just create profiles and then never do anything, and actively seeking them out to destroy them rarely achieves anything.

We already have rules in place on the system that prevents links in their profile from actually becoming links until they reach 10 reputation anyways. While some crawlers will still follow things that look like links (even in text), having them as plain text prevents our page rank or any kind of endorsement from passing along to it, and it's unlikely the crawler will actually index anything wherever it ends up.

So if you really feel like spending your time hunting these down, go for it, but by all means you should be absolutely certain it's an attempt at spam. If you can confidently make that assumption (because let's be honest, there are obvious cases out there), there's nothing particularly wrong with it. If there's even a hint that it might be a legitimate user, you should avoid doing it. As well, don't expect other moderators on other sites - or even us - to hop on board with you if you request a profile that's never done anything be nuked network-wide.

And before you run off, I ask you one question: Has doing this actually led to any noticeable decrease in the amount of spam making it onto the site? Most spammers who plan to create a post don't create a profile, leave it sit around long enough to be noticed, and then come back to post the spam. So I'm having a hard time imagining how this is helping with the original issue you mentioned was the "cause" of why you were doing this.

  • 3
    Has doing this actually led to any noticeable decrease in the amount of spam making it onto the site? - yes. On Puzzling, there was spam, before we started nuking them before they could post. After we started, I haven't seen any users last long enough to post the spam. – user58 Jul 4 '17 at 15:22
  • (and yes, via Smokey I have seen, more than once, users who had created an account and then came back years later to post obvious spam.) – user58 Jul 4 '17 at 15:27
  • 7
    @Mithrandir I'm... not seeing such a huge drop-off here. Looking at Dec '16, I'm seeing 11 posts spam/abusive deleted, and last month (Jun '17) there were 7 posts spam/abusive deleted. Doesn't seem like it's having that much of an effect. – animuson Jul 4 '17 at 15:28
  • I din't say it was a huge amount. But there was definitely the 'live stream' spam before we started preemptively nuking. But I haven't seen any after we started nuking, and neither has Smokey. – user58 Jul 4 '17 at 15:34
  • 8
    @Mithrandir It just seems like you're trying to convince yourself that something is working because you want it to be working, when in reality it's just using up a lot of effort for little gain. You should seriously consider analyzing the effectiveness of this strategy rather than just doing it. It might be a more worthwhile use of your time. – animuson Jul 4 '17 at 15:40
  • 1
    @mith If you're referring to what happened when Smokey was in its infancy - if the spam dropped off globally, while these nukes happened on puzzling only, there hardly could be any causal link. – John Dvorak Jul 4 '17 at 15:45
  • @JanDvorak - no, I meant in the past couple months. And on Puzzling only. – user58 Jul 4 '17 at 15:46
  • 2
    I have the same impression as Mithrandir about what has happened on Puzzling. On the other hand (1) I was actually the first Puzzling mod to start hammering spammers, so I may be vulnerable to the same cognitive bias animunson mentions, and (2) it's possible that the spam rate has changed but not for that reason (e.g., maybe it comes and goes with sporting seasons). – Gareth McCaughan Jul 4 '17 at 15:54
  • 3
    On the other other hand, while we're speaking of cognitive biases, it would be remiss not to point out that any mods on other sites who would prefer not to spend a lot of time hammering spammers have an incentive to convince themselves that it isn't doing any good :-). – Gareth McCaughan Jul 4 '17 at 15:55
  • 3
    @Mithrandir 11 to 7 is not statistically significant or, more accurately, your sample size isn't enough to make anything significant. – terdon Jul 4 '17 at 16:04
  • @terdon - I wasn't even talking about statistics, I was talking about the impression I got from personally seeing it. – user58 Jul 4 '17 at 16:05
  • @Mithrandir I know, my point is that you don't have enough data to form an impression. Not if we're talking of a change from 11 to 7. Your second comment about having seen users who came back to post spam years later is valid, but an impression based on so little data isn't really informative at all. – terdon Jul 4 '17 at 16:07
  • What about just created new users with 1. offensive usernames 2. profanity in the about me ... Mods can reset those (w/o them being flagged), right? – Bhargav Rao Jul 4 '17 at 16:27
  • @animunson If there are monthly figures for posts deleted as spammy/abusive, then presumably there's more than just those two numbers. 11 versus 7 tells us almost nothing, but (implausible purely-illustrative numbers) if it went 11,11,11,11,11,7,7,7,7,7 then that would be quite good evidence of a genuine effect. (Though, again, it could have some different cause.) Is there a way to get at those numbers? – Gareth McCaughan Jul 4 '17 at 16:29
  • @GarethMcCaughan I actually just used the search feature to find these posts. But I had to skip several months because a few of the recent ones had a bunch of abusive-deleted posts from a troll that created a bunch of profiles to post nonsense messages, and I didn't feel like digging that much to find how many were from normal flagging versus this one user causing problems. That's another problem with low numbers - a single person can throw them all off. – animuson Jul 4 '17 at 16:31
4

Here's an SEDE query that returns a list of obvious spammy profiles that serve exactly zero useful purpose. There are almost 1000 in the case of SO, with 11,000 links (half of which are in the first 200 profiles listed).

Here are links to the query on a few sites:

I haven't checked other sites. I didn't go looking for this data; I was working on something unrelated and this garbage got in the way. Frankly I've wasted enough time on these buggers, so do with this as you will.

Perhaps someone will find the list useful for testing (or better yet, nuking), or alternatively you us the list to fulfill all your fulfill hard-to-find shopping needs, from hookers to blow and Black Magic to Botox.

img

  • 2
    Nice effort! The counts are a bit high though, I spot-checked some of the results and - especially towards the bottoms of the lists - there were some that looked like real users who chose to enter links to places like their legitimate GitHub and Twitter accounts without any accompanying prose. I suppose it's an unavoidable consequence of just looking for links (it's not like you can do NLP from inside SEDE). – SOLO Jul 13 '18 at 13:50
1

What is the policy on destroying users with very spammy profiles but have not posted spam yet?

Productive use of time, "BeNice", and "Innocent until proven guilty" all play a role.

User ashleedawg posted a helpful answer with a SEDE tool that attempts to determine spammy accounts. Per the comments by Tensibai and Shadow Wizard links on the profile page and 'buying knives' is allowed.

Currently it's a case of waiting until the deed is done before meting out any punishment.

It may be some relief to know that the new SE Policy will delete inactive accounts.

As per the new data retention policy, when will inactive accounts now be deleted?

Blog - Changes to Our Privacy and Data Policies (May 21, 2018)

"part of our new data-retention policy, we will now delete accounts that have been inactive for more than two years and where the user has never engaged in any meaningful activity."

Meaningful Activity:

"Rather than holding on to that data forever, we plan to delete an account if a user:

  • hasn't visited the site in the last two years
  • and never asked or answered a question
  • and has a reputation of 1
  • and doesn't have a Developer Story or hasn't applied to a job or job search status ≠ actively looking
  • and hasn't opened or clicked any promotional email in the last year
  • and hasn't engaged in any other activity, such as edits, voting, bounties, etc.

As you can see, we are really targeting users who really, truly don't use our site."

Around May 2020 you might see many accounts deleted.

If that's too slow you might write those SE employees or answer/comment in that thread.

There is no flag for account profiles, but see here: How should I flag a user account if it has no posts to flag? - Perhaps a request for such a flag is useful but that is discussed here: Raise an automatic flag for spammers.

1

There is a difference between 'spammy' or 'potential spammers' accounts on the one hand and 'outright spam' accounts on the other hand.
If the information on the account is spam as the text stands, the account should be removed across the sites.

This is a sample of that kind of account:
https://stackexchange.com/users/1222778/loan-sydney?tab=accounts

  • 1
    Well, per-site mods can't really do anything about the other sites a spammer got account in. Only SE staff can handle those, I've brought this specific case in chat, let's see what the CM says about it. – Shadow Jul 17 '18 at 10:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .